Alassane BAH | Ecole Supérieure Polytechnique (original) (raw)

Papers by Alassane BAH

JASSS, 2024

Sahelian transhumance is a type of socio-economic and environmental pastoral mobility. It involve... more Sahelian transhumance is a type of socio-economic and environmental pastoral mobility. It involves the movement of herds from their terroir of origin (i.e., their original pastures) to one or more host terroir, followed by a return to the terroir of origin. According to certain pastoralists, the mobility of herds is planned to prevent environmental degradation, given the continuous dependence of these herds on their environment. However, these herds emit Greenhouse Gases (GHGs) in the areas they cross. Given that GHGs contribute to global warming, our long-term objective is to quantify the GHGs emitted by Sahelian herds. The determination of these herds' GHG emissions requires: (1) the artificial replication of the transhumance, and (2) precise knowledge of the space used during their transhumance. This article presents the design of an artificial replication of this transhumance through an agent-based model called MSTRANS. MSTRANS determines the space used by transhumant herds, based on the decision-making process of Sahelian transhumants. MSTRANS integrates a constrained multi-objective optimization problem and algorithms into an agent-based model. The constrained multi-objective optimization problem encapsulates the rationality and adaptability of pastoral strategies. Interactions between transhumants and their socio-economic network are modelled using algorithms and diffusion processes within the multi-objective optimization problem. The dynamics of pastoral resources are formalized at various spatio-temporal scales using equations that are integrated into the algorithms. The results of MSTRANS have been validated using GPS data collected from transhumant herds in Senegal. The MSTRANS results highlight the relevance of integrated models and constrained multi-objective optimization for modelling and monitoring the movement of transhumant herds in the Sahel. We can state that specialists in calculating greenhouse gas emissions now have a reproducible and reusable tool for determining the space occupied by transhumant herds in a Sahelian country. In addition, decision-makers, pastoralists, veterinarians and traders have a reproducible and reusable tool to help them make environmental and socio-economic decisions.

Free-range and common-grazing ruminants play a major role in the functioning and the provision of... more Free-range and common-grazing ruminants play a major role in the functioning and the provision of ecosystem services by agro-and silvo-pastoral ecosystems in sub-Saharan Africa. To assess how landscape structure affects the nature of environment-herd-services interactions, this paper describes a Multi-Agent System (MAS) that simulates daily herd movements in a dynamic environment. The model is used to study how the overall productivity of an agricultural territory is affected by the spatial organization of its different landscape units. The originality of this model lies in fine representation, in hourly time steps, of a herd's movements and activities over the seasons depending on changes in its environment. The herd is likened to a reactive agent with a global representation of its environment. It moves, grazes, drinks, rests and defecates. The model represents the spatio-temporal takeoff of biomass and animal faeces deposits. Grazing occurs between the moving, watering and resting phases, while defecation is continual. Biomass ingestion depends on its relative abundance in the explored plots, while the amount of faeces excreted depends on the amounts of biomass ingested the previous day. This paper describes the first use of the model on a choreme type environment representing a theoretical village territory typical of those found in West Africa. It comprises 5 landscape units: dwellings, compound fields, bush fields, fallows and rangelands. Two scenarios corresponding to 2 different landscape structures were simulated: alley fallows (Sc1), scattered fallows (Sc2). Most faeces deposits were in the fallows (sc1) or rangelands (sc2) in the wet season (4 out of 12 months) and in compound fields in the dry seasons (8 out of 12 months). Alley fallows (sc1) helped to intensify fertility transfers from rangelands to cultivated zones, thereby increasing productivity in cultivated zones (+26%) to the detriment of rangelands (-68%). The presence of ruminants in the territory therefore helped to i) enhance the positive effect of fallows and rangelands in renewing the fertility of cultivated soils and ii) maintain a system of concentric fertility rings with a decreasing fertility gradient from the core (=dwellings) to the periphery of the village (= rangelands). This modelling work thus helps to explain how landscape heterogeneity seen in African village territories is constructed.

L'Harmattan eBooks, 2017

Les ruminants divagants determinent le fonctionnement et la durabilite des ecosystemes agro et sy... more Les ruminants divagants determinent le fonctionnement et la durabilite des ecosystemes agro et sylvo-pastoraux. En vue d'etudier la dynamique de ces systemes complexes, ce chapitre s'interesse aux apports de la litterature internationale a propos de l'approche multi-agents pour modeliser le comportement spatial des animaux d'elevage. Partant de dix etudes couvrant une large diversite de contextes pedo-climatiques, de systemes modelises et de fonnalismes, une grille d'analyse est proposee afin de faciliter la construction des futurs modeles de simulation. Cette grille permet de mettre en lien les objectifs de modelisation avec le formalisme retenu pour representer l'environnement dans lequel evoluent les animaux, la distinction entre les differents agents qui constituent le troupeau, leurs capacites, et les regles qui guident les deplacements des animaux. Les auteurs recommandent aux futurs modelisateurs de mettre en coherence leur objectif de modelisation et le formalisme retenu. Le modele le plus complexe n'est pas necessairement le mieux adapte. Si l'objectif est d'etudier de facon theorique le comportement spatial et alimentaire des troupeaux, l'environnement peut etre represente de facon abstraite et simplifiee. Cependant, si l'objectif est d'etudier les interactions entre le troupeau et l'environnement, il est opportun de se referer a un environnement realiste via un SIG en distinguant par exemple la vegetation, les unites morphopedologiques et les amenagements. 1) Si l'accent est plutot mis sur l'effet des troupeaux sur l'environnement, le troupeau peut se reduire a un unique agent troupeau. Cet agent peut etre purement reactif ou cognitif. Dans le pennier cas, il est uniquement dote de capacites de representation et guide par des regles d'attraction/repulsion vis-avis des ressources et des amenagements. Dans le second cas, il a en plus des capacites de memorisation grâce a la presence d'un berger, on parle alors d'agent" troupeau-berger". 2) Si l'accent est plutot mis sur l'effet de l'environnement sur le troupeau, ce dernier peut etre represente de facon plus elaboree en distinguant un animal leader (agent cognitif) accompagne du reste du troupeau (agent reactif). Il est alors possible d'integrer des regles sociales entre animaux, de subordination entre le leader et le reste du troupeau en plus des regles d'attraction/repulsion. Ce formalisme permet par exemple de representer la variabilite alimentaire et donc de performance interanimale et intratroupeau. Distinguer en plus l'agent berger (agent cognitif) du troupeau permet d'etudier les strategies pastorales en representant les pratiques d'accompagnement. Cela permet, par exemple, d'evaluer l'interet d'une coordination entre bergers pour ameliorer la durabilite de l'ecosysteme agro ou sylvo-pastoral.

Marine Policy, Sep 1, 2018

Owing to high costs and difficulties in implement control, fisheries are less regulated than terr... more Owing to high costs and difficulties in implement control, fisheries are less regulated than terrestrial activities. In areas of weak governance, self-regulation is often the rule. Understanding the underlying bio-economic mechanisms that drive such fishery systems is crucial for decision making on how to support real fishermen communities. Fishing effort reallocation occurs when economic yield drops, suggesting a feedback between fishermen's mobility and local overexploitation. The nature of this feedback is addressed here by using a bioeconomic modelling approach applied to a Senegalese fishery. The white grouper is a highly valued commercial fish that has suffered intense exploitation in Senegal and was predicted to collapse due to a lack of fishery regulation combined with high demand. Surprisingly, our analysis of landings and associated price variability suggests that a slow recovery of this fishery is now underway. Motivated by this finding, two models were implemented that account for artisanal fishermen's mobility and variable prices. Both models provided the same insight: fishing effort mobility combined with variable fish prices induced a shift from local overexploitation to sub-regional sustainable exploitation. Considering that fishing effort mobility has steeply increased in recent years, this mechanism might have significantly contributed to the recent observed trends in prices and landings for this fishery. More generally, these bio-economic models suggest that spatial reallocation of fishing effort acts as emergent governance in under regulated areas that allows, in some cases, the recovery of an overexploited species.

International Journal of Biomathematics, Feb 1, 2019

A mathematical model for the dynamical transmission of polio is considered, with the aim of inves... more A mathematical model for the dynamical transmission of polio is considered, with the aim of investigating the impact of environment contamination. The model captures two infection pathways through both direct human-to-human transmission and indirect human-to-environment-to-human transmission by incorporating the environment as a transition and/or reservoir of viruses. We derive the basic reproduction number [Formula: see text]. We show that the disease free equilibrium is globally asymptotically stable (GAS) if [Formula: see text], while if [Formula: see text], there exists a unique endemic equilibrium which is locally asymptotically stable (LAS). Similar results hold for environmental contamination free sub-model (without the incorporation of the indirect transmission). At the endemic level, we show that the number of infected individuals for the model with the environmental-related contagion is greater than the corresponding number for the environmental contamination free sub-model. In conjunction with the inequality [Formula: see text], where [Formula: see text] is the basic reproduction number for the environmental contamination free sub-model, our finding suggests that the contaminated environment plays a detrimental role on the transmission dynamics of polio disease by increasing the endemic level and the severity of the outbreak. Therefore, it is natural to implement control strategies to reduce the severity of the disease by providing adequate hygienic living conditions, educate populations at risk to follow rigorously those basic hygienic rules in order to avoid adequate contacts with suspected contaminated objects. Further, we perform numerical simulations to support the theory.

Chaos, Dec 1, 2016

Neisseria meningitidis (Nm) is a major cause of bacterial meningitidis outbreaks in Africa and th... more Neisseria meningitidis (Nm) is a major cause of bacterial meningitidis outbreaks in Africa and the Middle East. The availability of yearly reported meningitis cases in the African meningitis belt offers the opportunity to analyze the transmission dynamics and the impact of control strategies. In this paper, we propose a method for the estimation of state variables that are not accessible to measurements and an unknown parameter in a Nm model. We suppose that the yearly number of Nm induced mortality and the total population are known inputs, which can be obtained from data, and the yearly number of new Nm cases is the model output. We also suppose that the Nm transmission rate is an unknown parameter. We first show how the recruitment rate into the population can be estimated using real data of the total population and Nm induced mortality. Then, we use an auxiliary system called observer whose solutions converge exponentially to those of the original model. This observer does not use the unknown infection transmission rate but only uses the known inputs and the model output. This allows us to estimate unmeasured state variables such as the number of carriers that play an important role in the transmission of the infection and the total number of infected individuals within a human community. Finally, we also provide a simple method to estimate the unknown Nm transmission rate. In order to validate the estimation results, numerical simulations are conducted using real data of Niger.

International Journal of Biomathematics, Jul 1, 2019

This paper has been motivated by the following biological question: how influential are desert ae... more This paper has been motivated by the following biological question: how influential are desert aerosols in the transmission of meningitidis serogroup A (MenA)? A mathematical model for the dynamical transmission of MenA is considered, with the aim of investigating the impact of desert aerosols. Sensitivity analysis of the model has been performed in order to determine the impact of related parameters on meningitis outbreak. We derive the basic reproduction number [Formula: see text]. We prove that there exists a threshold parameter [Formula: see text] such that when [Formula: see text], the disease-free equilibrium is globally asymptotically stable (GAS). However, when [Formula: see text], the model exhibits the phenomenon of backward bifurcation. At the endemic level, we show that the number of infectious individuals in the presence of desert aerosols is larger than the corresponding number without the presence of desert aerosols. In conjunction with the inequality [Formula: see text] where [Formula: see text] is the basic reproduction number without desert aerosols, we found that the ingestion of aerosols by carriers will increase the endemic level, and the severity of the outbreak. This suggests that the control of MenA passes through a combination of a large coverage vaccination of young susceptible individuals and the production of a vaccine with a high level of efficacy as well as respecting the hygienic rules to avoid the inhalation of desert aerosols. Theoretical results are supported by numerical simulations.

HAL (Le Centre pour la Communication Scientifique Directe), Jul 3, 2023

Les modèles de vision par ordinateur sont majoritairement entraînés avec de larges bases de donné... more Les modèles de vision par ordinateur sont majoritairement entraînés avec de larges bases de données publiques, dont l'exploration des biais permet l'identification de contextes dans lesquels ces modèles sont inadaptés. Le glissement d'un contexte adapté à un contexte inadapté est le témoin du manque de résilience d'un modèle à une variation dans les données. Nous tentons de caractériser le glissement de contexte géographique des données occidentales vers les données mondiales, et identifions les défis associés à cette caractérisation.

Springer eBooks, Sep 1, 2022

Sahelian transhumance is a seasonal pastoral mobility between the transhumant’s terroir of origin... more Sahelian transhumance is a seasonal pastoral mobility between the transhumant’s terroir of origin and one or more host terroirs. Sahelian transhumance can last several months and extend over hundreds of kilometers. Its purpose is to ensure efficient and inexpensive feeding of the herd’s ruminants. This paper describes an agent-based model to determine the spatio-temporal distribution of Sahelian transhumant herds and their impact on vegetation. Three scenarios based on different values of rainfall and the proportion of vegetation that can be grazed by transhumant herds are simulated. The results of the simulations show that the impact of Sahelian transhumant herds on vegetation is not significant and that rainfall does not impact the alley phase of transhumance. The beginning of the rainy season has a strong temporal impact on the spatial distribution of transhumant herds during the return phase of transhumance.

HAL (Le Centre pour la Communication Scientifique Directe), Jul 23, 2022

Popular benchmarks in Computer Vision suffer from a Western-centric bias that leads to a distribu... more Popular benchmarks in Computer Vision suffer from a Western-centric bias that leads to a distribution shift problem when trying to deploy Machine Learning systems in developing countries. Palliating this problem using the same data generation methods in poorly represented countries will likely bring the same bias that were initially observed. In this paper, we propose an adaptation of the MS COCO data generation methodology that address this issue, and show how the web scraping methods nests geographical distribution shifts.

L'Etat du Senegal a initie, depuis 1996, un processus de decentralisation visant la reappropr... more L'Etat du Senegal a initie, depuis 1996, un processus de decentralisation visant la reappropriation des politiques de developpement par le niveau local. Les collectivites locales issues de cette reforme ont acquis le pouvoir d'administrer, d'exploiter et de gerer les ressources naturelles et le foncier sous leur juridiction. L'Etat a ainsi transfere tout ou partie des attributions qu'il exercait au niveau central et territorial aux collectivites publiques et continue de les accompagner par ses services deconcentres, pour pallier le manque de moyens et de competences humaines des nouvelles entites administratives. Dans ce contexte, marque par l'insuffisance d'outils, de methodes et de documents de planification, la gestion des ressources naturelles et du foncier a l'echelle locale pose d'enormes difficultes aux elus charges du controle et de l'exercice du pouvoir. Aussi, pour reduire la fracture numerique et acceder a des informations pour la s...

Dans ce premier rapport, les principaux concepts SMA seront revisites et mis en perspectives. Dif... more Dans ce premier rapport, les principaux concepts SMA seront revisites et mis en perspectives. Differentes methodes de conception de systemes multi-agents seront egalement presentees. Il ne s'agira pas, dans cette partie, de discuter de la pertinence du choix de tel ou de tel autre type de modele. Ce travail doit nous permettre simplement de faire le point sur la demarche orientee agent qui est le type de modele choisi pour etudier la dynamique des ressources naturelles et des paysages a l'echelle du territoire dans le cadre du WP 5 du projet "Add Trans". (Resume d'auteur)

Globalisation of environmental, economic and social issues induces 28 rapid changes in agricultur... more Globalisation of environmental, economic and social issues induces 28 rapid changes in agricultural systems and land use. Competition or synergies 29 between different land users and farming systems are important issues. In this 30 context, the future of livestock farming systems (LS), which have often been 31 often described as having negative impacts on the environment, is debated. An 32 international comparison of processes of change in LS within seven territories 33 has been carried out to understand the driving factors and their impact on LS 34 dynamics and their relation to other land use management. As a common 35 paradigm, these studies used the participatory multi-agent modelling. A 36 synthesis of these models allowed describing a common conceptual model of 37 processes of change, to study scenarios and to compare the participatory 38 2 processes. Results show that common global drivers have different impacts on 39 livestock farming system at a local level but that LS are ...

Comme pour la premiere phase, l'execution de la deuxieme phase (2007-2009) du programme LEAD-... more Comme pour la premiere phase, l'execution de la deuxieme phase (2007-2009) du programme LEAD-SIPSA-FAO a ete confiee au PPZS. Elle avait pour objectif final de mettre en place un systeme d'aide a la decision pour anticiper, gerer et suivre les interactions entre le pastoralisme et l'environnement dans la zone sahelienne et contribuer a l'evolution des systemes pastoraux dans six pays du CILSS (Burkina, Mali, Mauritanie, Niger, Senegal et Tchad). Ce systeme regional d'information et d'alerte precoce doit a terme combler le manque de donnees, d'indicateurs et d'informations specifiques aux systemes pastoraux pour contribuer a la formulation de politiques. Au terme de deux ans et demi d'activites (ateliers nationaux, creation de comites nationaux de coordination, missions d'appui, collecte et elaboration de produits informationnels) la Coordination Technique Regionale (CTR) composee de l'URP PPZS, de FAO-LEAD et du Centre Regional AGRHYMET av...

Ecological Modelling, 2020

In this paper, we develop an agent-based model (ABM) to simulate the sylvatic transmission of the... more In this paper, we develop an agent-based model (ABM) to simulate the sylvatic transmission of the parasite T. cruzi, causative agent of Chagas disease (also called American trypanosomiasis). This ABM incorporates hostvector interactions, demographical processes for both hosts and vectors, and also vector and host dispersals. We implement the ABM under parameter values corresponding to real conditions in nature. The ABM simulations give quantitative results predicting the prevalence of the parasite T. cruzi in both hosts and vectors, the abundance of host and vector populations, and numerical estimates of the probability of persistence for T.cruzi population, depending on different initial spatial distributions of infection. Further, our study highlights the role of the host population as the determinant factor for infection transmission and parasite persistence, and presents pictures showing the emergence of clusters of infection.

Atmospheric and Climate Sciences, 2017

Guinea is confronted to the increasing risks of bushfires that destroy thousands of hectares of v... more Guinea is confronted to the increasing risks of bushfires that destroy thousands of hectares of vegetation cover every year. Very little research is devoted to the variability of those fires, which makes it a serious threat to both wildlife and human habitats. The current study investigates the spatial and temporal distribution of bushfires in the period from 2003 to 2016. The method used is the geospatial technology: we first filter pixels corresponding with active light supplied by MODIS images (Moderate Resolution Imaging Spectroradiometer) and estimate their densities following the square meshing procedure. Burned areas are deducted from the estimated pixel densities by calculations. The results highlight great occurrence of fires: 4 to 48 pixels of active fire per year and per 100 km² depending on the location; 2 to 5 million hectares per year of burned areas (20,000 to 50,000 sqkm). Almost 8 to 24% the size of the whole country. The prefectures of Beyla, Siguiri, Kouroussa, Kankan, Dinguiraye, Mali and Tougué are the most exposed areas. Every year, fire activities are observed as from October and between May and June. They are however mitigated according to the regions (or the geographical domains). Summits of bushfires activities are generally reached between December and January.